Undocumented text modes on Atari XL/XE

[Synthpopalooza]

Hi ...

 

I am new to this forum, and haven't actually programmed an 8-bit in about 20 years ... but have caught the bug again after playing around with Atari800Win.

 

Anyway, I had a few questions:

 

First off, during my time programming, I discovered 9 new text modes ... these are accessed using graphics mode 0, 1, and 2, and doing a poke to the GTIA chip (POKE 623). The following are the names I used:

 

GRAPHICS 0.1: 40x24 characters, 16 monochrome shades (sometimes called GRAPHICS 0/9)

GRAPHICS 0.2: 40x24, 10 programmable colours

GRAPHICS 0.3: 40x24, 16 hues (similar to GRAPHICS 11)

 

For these three modes, a 2x8 grid, with 4 bits per pixel, is used to design the characters.

 

GRAPHICS 1.1: 20x24 characters, about 9 monochrome shades.

GRAPHICS 1.2: 20x24, about 7 programmable colours.

GRAPHICS 1.3: 20x24, about 9 hues

 

For these three modes, a 4x8 grid with 2 bits per pixel (same as the ANTIC 4/5 modes) is used to design the characters. The colours change slightly when you print lowercase and lowercase/inverse characters to the screen.

 

GRAPHICS 2.1: 20x12, 9 monochrome shades

GRAPHICS 2.2: 20x12, 7 colours

GRAPHICS 2.3: 20x12, 9 hues

 

Same as above, but lower resolution.

 

To access these modes, you do the appropriate GRAPHICS command, then POKE 623, gtiamode*64. You could also set up a display list interrupt using the gtia register.

 

I was wondering if anyone could document exactly what colours get used in GRAPHICS 1 and 2, when when you use GTIA. I've also been curious about the feasibility of using GRAPHICS 13 (antic 5) in combination with the GTIA registers. In this one, you get the 2x8 grid again, but with fewer colours. However, it uses less memory than GRAPHICS 0.

 

I did actually write a Character design program which uses these GTIA modes, unfortunately it's stuck on a bunch of 5.25" floppies with no way for me to access them at present.

[Rybags]

They're GTIA modes, equivalent to 9, 10, 11 (your PRIOR values 64, 128, 192).

 

Visually, they work the same and the bit patterns in the character set choose colours in the same way as if you had bitmap data in Mode F.

Except if you have Inverse characters (high bit set). In that case, the bit patterns are inverted.

 

You can set GTIA modes with practically any Antic mode, but in most cases they're not useful as you can't select the full possible colour range.

 

In Graphics 1, 2 it's not useful since those modes only allow 2 colours per character cell.

 

In Graphics 12, 13 (Antic 4, 5) they're partially useful. You can't generate all of the possible ANx bus combinations within a character cell, so you can't get all 16 colours or luma values in a single character cell.

To be specific, the high bit of the character decides whether Playfield 2 or 3 is used for bit pattern "11". That equates to a bit pair on the ANx bus of "10" or "11".

[Synthpopalooza]

You can set GTIA modes with practically any Antic mode, but in most cases they're not useful as you can't select the full possible colour range.

 

In Graphics 1, 2 it's not useful since those modes only allow 2 colours per character cell.

 

In Graphics 12, 13 (Antic 4, 5) they're partially useful. You can't generate all of the possible ANx bus combinations within a character cell, so you can't get all 16 colours or luma values in a single character cell.

To be specific, the high bit of the character decides whether Playfield 2 or 3 is used for bit pattern "11". That equates to a bit pair on the ANx bus of "10" or "11".

 

I have actually gotten some practical use out of modes 1 and 2 with the GTIA set. In the lower atascii range each character can have the bit set to one of four colours, much like ANTIC 4 ... higher characters only allow 2 colours.

 

And I found Graphics 12 (antic 4) with the GTIA bit set hardly useful, because for the amount of memory it takes, you can just use GRAPHICS 0 and have access to the full 16 colours, at the same resolution.

[Rybags]

Antic 4 does have one good use - for doing CIN type imagery, ie alternating scanlines of Colour mode (PRIOR=C0) and normal mode.

 

That gives you (with limitations) the ability to do 16 colour * 4 luma imagery.

 

You can get the same effect in Mode 2 though, although not all emulator versions handle it properly. By setting PRIOR to a GTIA mode then back to 00 during a scanline, you get the mode where 4 colours are available in 160 across, with pixel values mapped to PF0 thru PF3.

Of course doing either of those involves a Kernal that uses 100% CPU in the screen region used.

[NRV]

Of course doing either of those involves a Kernal that uses 100% CPU in the screen region used.

 

Not really.. just the 50% or less..

I have done this with IRQ's but I suppose you can do it with DLI's also..

For the GTIA 9+11 mode you can set an IRQ to happen every two scanlines, at the very start of the interruption you set the PRIOR value for mode 11, then waste some cycles and when the scan is outside the screen you set PRIOR to the value for mode 9, then exit your interruption.. so you don't need to do anything in the next scanline

I did this for narrow mode and I think I can do it for normal width, but not for the wide mode, because there are cases when the interruption starts too late

Also I have tested this only in the emulator

 

Regards

 

NRV

[Rybags]

I guess you could use Timers but then you'd probably want to do it via a RAM-based OS with the IRQ vector at $FFFE.

You also then have the problem that in emulation, Timers don't necessarily work properly, ie they won't count in cycle-exact fashion.

[NRV]

I guess you could use Timers but then you'd probably want to do it via a RAM-based OS with the IRQ vector at $FFFE.

You also then have the problem that in emulation, Timers don't necessarily work properly, ie they won't count in cycle-exact fashion.

 

Yep, that was what I did, you need to use the "hardware" vectors, if not there is no enough time.. I should test it in real hardware, because if the IRQ's doesn't start at the same point, in the emulation and in a real computer, this method could not work..

 

Anyways, you can also do it with DLI's (at least the GTIA 9+11 mode), but only for narrow mode and with a "high" cost

I needed to use the NMI hardware vector AND lost the "x" index (i cannot use it in the main code, only for the interruption code), because the first instruction of the interruption is always "STX PRIOR".. with this you can be sure that PRIOR has the correct value before the start of the line (and don't need to use WSYNC)

 

This is the code for the "one DLI per scanline" method. You can do also one DLI every two scanlines, I never did because I started using IRQ's.. I wanted my "X" back

 

;----------------------------------------
; display list interrupt 1 code
;----------------------------------------

DLI1_address
stx PRIOR
ldx #PRV_GTIA_9

;----------------------------------------
sta m_dliSaveA

lda #<DLI2_address
sta NMIH_VECTOR

lda m_dliSaveA

rti


;----------------------------------------
; display list interrupt 2 code
;----------------------------------------

DLI2_address
stx PRIOR
ldx #PRV_GTIA_11

;----------------------------------------
sta m_dliSaveA

lda #<DLI1_address
sta NMIH_VECTOR

dec m_dliLineCounter
beq DLI2_last_line

lda m_dliSaveA

rti

;----------------------------------------
DLI2_last_line
lda #MAX_DLI_LINE_CTD
sta m_dliLineCounter	; reset last line counter

lda #0			; reset PRIOR to 0 for the rest of the screen
sta WSYNC
sta PRIOR

lda m_dliSaveA

rti

 

NRV

[Synthpopalooza]

I've documented the GTIA settings for Graphics modes 1 and 2, along with the colors used.

 

Note that in these modes, ATASCII 0-63 will not print at all, all bit combinations in the 4x8 grid produce the background color.

 

These modes use 3 palettes, and a 4x8 grid like ANTIC 4/5. The different palletes are accessed in the different ATASCII ranges. Normal graphics 1 (GTIA 0) is ATASCII 0-63 register 708, 64-127 register 709, 128-191, register 710, 192-255 register 711.

 

graphics 1/10,2/10

 

BITS

__________________

00 01 10 11 Graphics 10 colors

 

64-127: 704, 705, 708, 709 (0, 1, 4, 5)

128-191: 704, 706, 712, 712 (0, 2, 8 )

192-255: 704, 707, 708, 711 (0, 3, 4, 7)

 

8 possible colors, 704 is the background.

 

 

graphics 1/9,2/9

 

BITS

__________________

00 01 10 11

 

64-127: 0 1 4 5

128-191: 0 2 9 10

192-255: 0 3 14 15

 

10 monochrome shades of one color

 

 

graphics 1/11,2/11

 

BITS

__________________

00 01 10 11 Graphics 11 colors

 

64-127: 0 1 4 5 (black, yellow, purple, indigo)

128-191: 0 2 9 10 (black, orange, blue, teal)

192-255: 0 3 14 15 (black, red, olive, gold)

 

10 colors of the same luminance

 

ANTIC 4 character sets work perfectly in here, except that the second pallete in 1/10, 2/10 has only 3 colors available. Bits 10 and 11 correspond to register 712.

 

It would be interesting to see how the software super IRG (GemDrop) mode would work, with these GTIA modes used instead of ANTIC 4!

Edited July 8, 2009 by Synthpopalooza

[Synthpopalooza]

Just discovered something else ...

 

I was reading an article where someone had interlaced Graphics 10 with Graphics 9 and gotten a 160x192 resolution, because of a bug which shifts the Graphics 10 screen right by 1/2 pixel.

 

Well, I tried this little trick in Atari TurboBASIC in Graphics 0:

 

DO:POKE 623,128:POKE 623,64:LOOP

 

The character grid gets shifted right by 1/2 a gtia pixel when it cycles into GTIA 2 (graphics 10).

 

What this means is you can apply Bill Kendrick's Super-IRG shift and get a 4x8 antic 4 style grid, with at least 160 possible colours! I am going to reverse engineer his font designer and see if I can design some demo fonts!

 

In Graphics 1, you can get a full 8x8 character grid, with (possibly) 70 colours! 16 displayable per character grid.

 

This trick also works if you shift mode 10 with mode 11.

Edited July 10, 2009 by Synthpopalooza

[Rybags]

That's old news.

 

Known around the place as HIP, RIP and TIP modes.

 

HIP uses Gr. 9 and 10 interleaved, 16 luma and 8 luma in monochrome. And you alternate the sequence per frame.

 

TIP uses Gr. 11 then 9, 11 then 10 (then repeats). Alternate per frame for perceived 160 pixels across.

 

I don't get what you're talking about with Graphics 1... that mode is pretty useless for GTIA modes because you can't generate many bit combinations.

[José Pereira]

Hello.

Can someone of you post some screen examples of this modes?

Or Emulator files to download?

 

 

Thanks.

José Pereira.

[Synthpopalooza]

That's old news.

 

Known around the place as HIP, RIP and TIP modes.

 

HIP uses Gr. 9 and 10 interleaved, 16 luma and 8 luma in monochrome. And you alternate the sequence per frame.

 

TIP uses Gr. 11 then 9, 11 then 10 (then repeats). Alternate per frame for perceived 160 pixels across.

 

I am aware of this. I am not talking about the HIP and TIP modes, I am talking about using the GTIA bug (documented in 1996) and applying it to a bitmap mode (Graphics 0, 1, and 2) with the GTIA bit set, and cycling between them in a manner similar to Bill Kendrick's Super IRG mode.

 

I don't get what you're talking about with Graphics 1... that mode is pretty useless for GTIA modes because you can't generate many bit combinations.

 

You can generate 10 hues/monochrome shades if you set your GTIA to mode 9/11 in Graphics 1, and 8 programmable colors (all except for register 710) if you set the GTIA to mode 10 in Graphics 1.

 

You get a 4x8 ANTIC 4 style character grid, with 3 distinct color palettes that these colours are spread across. The three ATASCII ranges past 64 (i.e. lowercase, inverse, and inverse lowercase) access these palettes. The only real flaw here, is that ATASCII 0-63 are nondisplayable.

 

If someone can point me to a good multicolor ANTIC 4 font, I can demonstrate this with a screenshot, but I know this can be done! Meanwhile I am posting these screenshots using the normal Atari font (they are at the bottom of this post)

 

 

Picture 1: This is normal Graphics 1, with ATASCII 0-255 printed. Each range on each line. This we know well.

 

 

Picture 2: Graphics 1 GTIA 1 (mode 9). ATASCII 0-63 are undisplayable, but if you look closely you will see 4 bit combinations unique to each row of characters. This produces 10 shades of one monochrome colour. Specifically:

 

ATASCII 0-63: All bit pair combinations generate the background colour.

ATASCII 64-127: (using Graphics 9 colors): 00-0 01-1 10-4 11-5

ATASCII 128-191: 00-0 01-2 10-9 11-10

ATASCII 192-255: 00-0 01-3 10-14 11-15

 

 

Picture 3: graphics 1 with gtia 2 (mode 10)

 

In this image, I have done POKE 704,0 and POKE 712,15, plus POKES to 705-707 to activate those colours.

 

Here, you will see 4 unique bit pair combinations in lines 2 and 4, and three unique bit pair combinations in line 3. This produces 8 programmable colors (only register 710 cannot be shown here).

 

ATASCII 64-127: 00-704 01-705 10-708 11-709 (corresponding to mode 10 colors 0, 1, 4, 5)

ATASCII 128-191: 00-704 01-706 10-712 11-712 (3 colour palette - mode 10 colors 0, 2, 8 )

ATASCII 192-255: 00-704 01-707 10-708 11-711 (corresponding to mode 10 colors 0, 3, 4, 7)

 

You will notice that register 708 (Graphics 10 color 4) is shared between ATASCII-64-127 and ATASCII 192-255. You will also notice that, like in mode 10, 704 becomes the background colour, so care would need to be taken when using player-missile graphics.

 

 

Picture 4: Graphics 1 with GTIA 3 (mode 11)

 

Similar to mode 9 only with hues, instead of shades. Again, 4 unique bit combinations per line, for 10 colours.

 

ATASCII 64-127: (using Graphics 11 colors): 00-0 01-1 (yellow) 10-4 (purple) 11-5 (indigo)

ATASCII 128-191: 00-0 01-2 (orange) 10-9 (blue) 11-10 (teal)

ATASCII 192-255: 00-0 01-3 (red) 10-14 (olive) 11-15 (gold)

 

Now, here's what I am talking about:

 

Firstly, using Graphics 1/10 in a manner similar to Bill Kendrick's Super IRG can potentially generate 22 displayable colors in this mode. You'd use one whole character set (256 chars) and strobe register 756 between each half during VBLANK.

 

 

However, if you toggle between mode 10 and one of the other modes, like in the HIP and TIP modes, but from within graphics 1, the text onscreen shifts 1/2 pixel to the right. Combine this with a character set shift at register 756 (remember, in mode 1 the last half of the character set is not displayed unless you change 756) and you have an effective interlacing 8x8 character grid, with a possible 12-16 colors displayable per grid. The other requirement is of course shifting register 712 to the background color when you strobe into mode 9. This gets you a total of potentially 44 displayable interlaced colors in this mode.

 

Now, if you were to try this in GRAPHICS 0, where you get a 2x8 character grid with the full range of GTIA colors, and use a method similar to Bill Kendrick's SUPER IRG, the interlacing character grids would produce a 4x8 (ANTIC 4) grid, but with potentially up to 144 colours. (9 mode 10 * 16 mode 9) if you interlace mode 10 with mode 9. The limits of this, I don't know yet, but I think it has great potential. Here though, you would need 2 whole character sets, as with Super IRG, and shift between them during the interlace operation. And again, set register 712 to zero when you shift into mode 9.

Edited July 10, 2009 by Synthpopalooza

[Rybags]

I should let you know... don't trust the emulator. It doesn't necessarily generate GTIA modes properly for all Antic modes.

 

Toggling PRIOR between 64/128 in Graphics 0 is no different to toggling between a Gr. 9 and Gr. 10 screen, so far as the effect.

 

Don't want to burst your bubble here, but you've not really uncovered anything new.

 

And pallette swapping per VBlank - been done before. Not really a fan of it here, you just get annoying flicker.

[Synthpopalooza]

I should let you know... don't trust the emulator. It doesn't necessarily generate GTIA modes properly for all Antic modes.

 

Perhaps, but I have tested this on a real Atari as well, with the same results. Admittedly this was years ago when my Atari was actually working, but I did even manage to write some programs (including a character set editor) using these modes.

 

Can someone with a real Atari confirm this?

[Rybags]

Well, GTIA modes work in Gr. 1 but...

 

they're not very useful. Like you said, Playfield 0 characters will never display. That is because they equate to "00" on the AN0-1 lines from Antic.

So, the end result is that in GTIA modes in Gr. 1, you will have maximum of 3 colours inside a character cell, with fat pixels.

 

Sure, you get more colours available than normal but at half the resolution. You're way better off just sticking to Graphics 0 with GTIA modes because you have much less restrictions and can use the modes in virtually the same manner as your normal Gr. 9, 10, 11.

[Synthpopalooza]

Well, GTIA modes work in Gr. 1 but...

 

they're not very useful. Like you said, Playfield 0 characters will never display. That is because they equate to "00" on the AN0-1 lines from Antic.

So, the end result is that in GTIA modes in Gr. 1, you will have maximum of 3 colours inside a character cell, with fat pixels.

 

Sure, you get more colours available than normal but at half the resolution. You're way better off just sticking to Graphics 0 with GTIA modes because you have much less restrictions and can use the modes in virtually the same manner as your normal Gr. 9, 10, 11.

 

True enough. The advantage to using mode 1 is, of course, less memory required. It's a tradeoff, less colours vs more available RAM. It depends on how important conserving RAM is to you versus how many colours you want to display at once.

[José Pereira]

Hi ...

 

I am new to this forum, and haven't actually programmed an 8-bit in about 20 years ... but have caught the bug again after playing around with Atari800Win.

 

Anyway, I had a few questions:

 

First off, during my time programming, I discovered 9 new text modes ... these are accessed using graphics mode 0, 1, and 2, and doing a poke to the GTIA chip (POKE 623). The following are the names I used:

 

GRAPHICS 0.1: 40x24 characters, 16 monochrome shades (sometimes called GRAPHICS 0/9)

GRAPHICS 0.2: 40x24, 10 programmable colours

GRAPHICS 0.3: 40x24, 16 hues (similar to GRAPHICS 11)

 

For these three modes, a 2x8 grid, with 4 bits per pixel, is used to design the characters.

 

GRAPHICS 1.1: 20x24 characters, about 9 monochrome shades.

GRAPHICS 1.2: 20x24, about 7 programmable colours.

GRAPHICS 1.3: 20x24, about 9 hues

 

For these three modes, a 4x8 grid with 2 bits per pixel (same as the ANTIC 4/5 modes) is used to design the characters. The colours change slightly when you print lowercase and lowercase/inverse characters to the screen.

 

GRAPHICS 2.1: 20x12, 9 monochrome shades

GRAPHICS 2.2: 20x12, 7 colours

GRAPHICS 2.3: 20x12, 9 hues

 

Same as above, but lower resolution.

 

To access these modes, you do the appropriate GRAPHICS command, then POKE 623, gtiamode*64. You could also set up a display list interrupt using the gtia register.

 

I was wondering if anyone could document exactly what colours get used in GRAPHICS 1 and 2, when when you use GTIA. I've also been curious about the feasibility of using GRAPHICS 13 (antic 5) in combination with the GTIA registers. In this one, you get the 2x8 grid again, but with fewer colours. However, it uses less memory than GRAPHICS 0.

 

I did actually write a Character design program which uses these GTIA modes, unfortunately it's stuck on a bunch of 5.25" floppies with no way for me to access them at present.

 

 

Hi, where's Synth...?

(last active at February,13)

 

Is there any Demos, Editor, YouTube, Slides,... of Modes 1.1, 1.2 and 1.3

(not Synth...'s, from others, even on the past?)

 

 

Another one I would like to see more tries of pictures and running is Super-IRG.

(some tries that get the less flicker as possible)

 

 

Just this one more:

What is that Psycho GTIA Mode at the MadsTeam Web Page?

It is a canceled Project, but I am with difficulties to understand even using Portuguese or English translations... is this what? An add on, or just a programming trick (like GTIA Bug). It seems very interesting. That GTIA pixel shows in one of the 2:1 registers (for me it seems Backgr.-Black, is it?). Very interesting, because it will possible add 16 colour/Luminances to 4PFs. and probably leave PMs. free...

Anyone could xplain this one in a better English, please?

And more screenss, files or any Videos running this Mode?

 

 

Greetings.

José Pereira.

[José Pereira]

Another one, probably using GTIA Bug, but not the traditional one: ILLUSIA Demo at YouTube:

It is 2:1 than a Window with 4:1 running and returns to 2:1.

 

Technical notes»__________________________?

 

Thanks.

José Pereira.

[Rybags]

How about some links to these things you're talking about. I have no idea what "Illusia" is, and I don't know where the Madteam homepage is.

[José Pereira]

How about some links to these things you're talking about. I have no idea what "Illusia" is, and I don't know where the Madteam homepage is.

 

 

Sorry.

 

Ilusia on YouTube: www.youtube.com/watch?v=1w39gQt9v3k

 

And about MadTeam, just visit their page and it's almost on the Bottom of the left side choose panel, called Psycho GTIA: www.madteam.atari8.info

 

José Pereira.

[MaPa]

How about some links to these things you're talking about. I have no idea what "Illusia" is, and I don't know where the Madteam homepage is.

 

Madteam homepage - http://madteam.atari8.info/

 

You will find the demo there too (http://madteam.atari8.info/index.php?prod=scena), bottom of the page. Use real HW or Altirra emulator as it doesn't run properly in Atari800Win.

Edited June 22, 2010 by MaPa

[Rybags]

The Psycho "mode" in fact seems to be some hardware mod that uses a processor that can alter GTIA registers at high speed. From the translation it seems it might have ended up costing more to make than a VBXE.

 

The Illusia demo just changes to GTIA mode and back - been done before plenty of times but I don't know that any other demo had moving graphics like it does.

 

"SuperIRG"... not so super IMO. The concept of changing between colours each VBlank is old and rarely used because it just doesn't look so great.

[José Pereira]

Just discovered something else ...

 

I was reading an article where someone had interlaced Graphics 10 with Graphics 9 and gotten a 160x192 resolution, because of a bug which shifts the Graphics 10 screen right by 1/2 pixel.

 

Well, I tried this little trick in Atari TurboBASIC in Graphics 0:

 

DO:POKE 623,128:POKE 623,64:LOOP

 

The character grid gets shifted right by 1/2 a gtia pixel when it cycles into GTIA 2 (graphics 10).

 

What this means is you can apply Bill Kendrick's Super-IRG shift and get a 4x8 antic 4 style grid, with at least 160 possible colours! I am going to reverse engineer his font designer and see if I can design some demo fonts!

 

In Graphics 1, you can get a full 8x8 character grid, with (possibly) 70 colours! 16 displayable per character grid.

 

 

This trick also works if you shift mode 10 with mode 11.

 

 

The 1/2 shifted turn 4:1 into 2:1 and lots of colours.

Any real documentation/information, Type-in or Screen examples?

 

Any one knows or ever tried this one?

 

 

 

Thanks.

Greetings.

José Pereira.

[Rybags]

Been done. HIP, TIP and RIP modes use the Gr. 10 shifted pixel to simulate 160 across instead of 80.

 

As much as I don't much like to say it - there is probably little to nothing left to "discover" insofar as non documented tricks or features of Atari text modes, or bitmap modes either for that matter.

 

The only trick remaining to find, if it exists, is how to get Gr. 9 or 11 pixels to shift the 1/2 pixel, although it's thought to only occur in certain GTIA batches.

[José Pereira]

Been done. HIP, TIP and RIP modes use the Gr. 10 shifted pixel to simulate 160 across instead of 80.

 

As much as I don't much like to say it - there is probably little to nothing left to "discover" insofar as non documented tricks or features of Atari text modes, or bitmap modes either for that matter.

 

The only trick remaining to find, if it exists, is how to get Gr. 9 or 11 pixels to shift the 1/2 pixel, although it's thought to only occur in certain GTIA batches.

 

 

POKE 623,128:POKE 623,64

 

This puts Gr.10 and than 9.

 

From what you're saying it will be what?

POKE 623,64:POKE 623,192

 

 

 

And then: plot pixels across screen with Setcolor, color,...

But, if so, what I must have?

 

Gr.9:

SETCOLOR 4,1,0 (select colour1)

COLOR MODE9 (MODE9=0TO15_Luminances)

PLOT ....,....

 

Gr.11:

SETCOLOR 4,0,6 (Select Lumin.=6)

COLOR MODE11 (select colour)

PLOT ....,....

 

 

It is probably more...

Cannot acess Emulators to try right now...

 

 

 

But if they try the ones you say, why no-one try this one?

 

 

 

 

Thanks.

Greetings.

José Pereira.